In algebraic topology, we have the map lifting lemma which says that given a covering space $p:(\tilde{X},\tilde{x})\rightarrow (X,x)$ and a map $f:(Y,y)\rightarrow (X,x)$ with $Y$ connected and locally path-connected, a lift $\tilde{f}:(Y,y)\rightarrow(\tilde{X},\tilde{x})$ of $f$ exists if and only if $f_*(\pi_1(Y,y))\subset p_*(\pi_1(\tilde{X},\tilde{x}))$.

I want to ask whether analogues of the above are true for finite etale maps and etale fundamental groups.


The answer is yes, at least for finite covers (or even pro-finite covers). The key is to carefully translate everything into the language of finite $\pi_1$-sets using the Galois correspondence. The canonical reference is of course SGA 1, Expose V - in particular $\S5,6$ and 7 once you're familiar with the basics.

By the universal property of fiber products, the existence of such a lifting $\tilde{f}$ is equivalent to the existence of a section of the covering map (ie finite etale morphism) $p_Y : Y\times_X\tilde{X}\rightarrow Y$ which contains the geometric point $(y,\tilde{x})$ (such a section if it exists, is unique).

By Galois theory, the pointed covering $p : (\tilde{X},\tilde{x})\rightarrow (X,x)$ is equivalent to the data of the pointed set $(p^{-1}(x),\tilde{x})$, together with the monodromy action of $\pi_1(X,x)$ on the underlying set $p^{-1}(x)$. The pullback $p_Y$ is (via the Galois correspondence) given by the pointed set $(p_Y^{-1}(y),(y,\tilde{x}))$ together with the monodromy action of $\pi_1(Y,y)$ acting on the underlying set, which is described by the map $$\pi_1(Y,y)\stackrel{f_*}{\rightarrow}\pi_1(X,x)\rightarrow Aut(p^{-1}(x))\cong Aut(p_Y^{-1}(y))\qquad (*)$$ where the last isomorphism is induced by the natural bijection $p_Y^{-1}(y)\cong p^{-1}(x)$.

(In short, the pullback of a cover over $X$ given by a $\pi_1(X)$-set $T$ by any morphism $Y\rightarrow X$ is the cover over $Y$ given by pulling back the $\pi$-set structure on $T$ via the morphism $\pi_1(Y)\rightarrow\pi_1(X)$. (What else could it be?) See $\S6$ in SGA 1 Expose V)

Under these identifications, $\pi_1(\tilde{X},\tilde{x})$ is naturally identified with the subgroup of $\pi_1(X,x)$ defined by $Stab_{\pi_1(X,x)}(\tilde{x})$. The map $p_Y$ admits a section containing the geometric point $(y,\tilde{x})$ if and only if the point $(y,\tilde{x})\in p_Y^{-1}(y)$ is fixed by the monodromy action of $\pi_1(Y,y)$ (Note that a section of $p_Y$ is a morphism of finite etale covers over $Y$, and hence corresponds via the Galois correspondence to a $\pi_1(Y,y)$-set morphism from the singleton set with trivial $\pi$-action to $p_Y^{-1}(y)$. Since this morphism is a morphism of $\pi_1(Y,y)$-sets, and hence must commute with the $\pi_1(Y,y)$-action, it implies that the image of this map is fixed by $\pi_1(Y,y)$).

However, by the definition of this monodromy action in $(*)$, we see that this is true if and only if $f_*\pi_1(Y,y)\subset Stab_{\pi_1(X,x)}(\tilde{x}) = \pi_1(\tilde{X},\tilde{x})$.

| cite | improve this answer | |
  • 3
    $\begingroup$ If I may: note that this gives a rather nice proof of the original statement in the topological case. $\endgroup$ – Piotr Achinger Jun 11 '18 at 19:29
  • $\begingroup$ @PiotrAchinger Hi Piotr! I guess I'm a little worried about the situation for infinite covers. Does the Galois correspondence (equiv. of categories between finite covers and finite pi-sets) extend to arbitrary covers? (iirc a Galois category as defined by grothendieck must have a fundamental functor taking values in finite sets) $\endgroup$ – Will Chen Jun 19 '18 at 13:12

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.